PDA

View Full Version : stream.mp3 hiccups



herbman
2010-02-24, 07:29
I have a situation on my system (7.4.2, athlon xp 1700+, 512M RAM, 4x250G SATA RAID-5) where I get skips when I run the stream.mp3 stream from work to my itunes player. I have absolutely no issues at home with my slimp3 (so the server is fine), and I know for sure that it's not a bandwidth/network issue. The skip I describe is that itunes (streaming buffer set to large) has to rebuffer the stream every 15-30 seconds or so.

When the stream is playing, the only piece of real evidence I have is that it skips only when going through the transcoding pipeline to match my desired bandwidth cap of 128. I've used the stock convert.conf, and also modified it to use cbr 128 instead of vbr, with no difference.

I turned on the 'transcoder' set of logs, and it seems like this pattern is in place when I find a skip:
[10-02-24 09:28:19.0794] Slim::Player::Pipeline::sysread (301) Attempting to write to pipeline writer
[10-02-24 09:28:19.0802] Slim::Player::Pipeline::sysread (307) Wrote 4096 bytes to pipeline writer
[10-02-24 09:28:19.0809] Slim::Player::Pipeline::sysread (301) Attempting to write to pipeline writer
[10-02-24 09:28:19.0827] Slim::Player::Pipeline::sysread (301) Attempting to write to pipeline writer
[10-02-24 09:28:19.4857] Slim::Player::Pipeline::sysread (301) Attempting to write to pipeline writer
[10-02-24 09:28:19.4865] Slim::Player::Pipeline::sysread (307) Wrote 4096 bytes to pipeline writer
[10-02-24 09:28:19.4871] Slim::Player::Pipeline::sysread (279) Pipeline doesn't have pending bytes - trying to get some from source
[10-02-24 09:28:19.4881] Slim::Player::Pipeline::sysread (301) Attempting to write to pipeline writer
[10-02-24 09:28:19.4903] Slim::Player::Pipeline::sysread (307) Wrote 4096 bytes to pipeline writer

Any ideas here on how to further diagnose? I have a self-compiled lame binary (3.95 MMX) on fedora core 3. Is it possible that my cpu is too slow to properly feed the transcoder while converting mp3 to mp3? Or that my disk subsystem is too slow to feed the data (i doubt it, it's just mp3)..

Seeking ideas for diagnosis.